Acid3: NodeIterator fix leaves us with 4 points to go
Posted by Darin Adler on Tuesday, March 25th, 2008 at 5:49 pmWebKit was still failing on test 2 of Acid3. That’s a NodeIterator test. The problem was a simple one: While NodeIterator objects were rigged to update themselves when nodes are removed from the document, there were a couple of edge cases that weren’t handled quite right. I tried to fix this about a month ago, but got confused. This time I was able to keep my head straight and it took me less than an hour to fix the bug.
We’re passing 96/100 tests now. There are only 4 points to go to get to 100 points. There are also a few loose ends that don’t cost a point and also some reports of a few failures that some see on their computers, but don’t affect most computers.
We’re closing in on 100%!
March 25th, 2008 at 6:33 pm
when the test says that it must be pixel perfect, does that mean that font rendering is fully specified to the pixel?
March 25th, 2008 at 6:36 pm
@masterleep: If you look at the reference rendering of Acid 3 you’ll notice that it is made up of a background image with the text positioned over top of it via CSS. By having the browser renders its own text for the reference rendering it is possible to expect that the test and reference match exactly. If the text were part of the reference image, then there would obviously be variance due to differing operating systems, font smoothing settings, and so forth.
March 25th, 2008 at 6:56 pm
ah, i expected that the reference rendering was an image. Thanks for the explanation.
March 25th, 2008 at 7:02 pm
There is certainly some difference with my browser and the thickness of the borders around the white box and the internal coloured boxes. Acid3 rendering with WebKit is slightly smaller than in the reference rendering.
March 26th, 2008 at 1:42 am
And so, when webkit will reach 100%, we’ll get Safari 3.2?
March 26th, 2008 at 1:47 am
I figure Apple will be very quick to get a stable update to Safari released so they can claim to be the first browser to be Acid3 compliant, and by a large margin.
March 26th, 2008 at 3:22 am
What about the upper right corner? Seems to me that there’s more than a few pixel to fix there
Well done however, keep up with the good work!
March 26th, 2008 at 3:36 am
Don’t know if you’ve noticed, but the font smoothing of ‘Acid3′ differs between reference (subpixel) and test page (grayscale).
March 26th, 2008 at 4:48 am
Have you fixed some other timing-bugs? I now consistently get 96/100, even after pagereload etc. In the last public build (which was supposed to get 95/100) I got everything from 90-95, different on each pageload. Keep up the _great_ work!
March 26th, 2008 at 12:55 pm
Patrick, can you file a bug report about the issue to http://bugs.webkit.org?